In this notebook, some template code has already been provided for you, and you will need to implement additional functionality to successfully complete this project. You will not need to modify the included code beyond what is requested. Sections that begin with '(IMPLEMENTATION)' in the header indicate that the following block of code will require additional functionality which you must provide. Instructions will be provided for each section, and the specifics of the implementation are marked in the code block with a 'TODO' statement. Please be sure to read the instructions carefully!
Note: Once you have completed all of the code implementations, you need to finalize your work by exporting the iPython Notebook as an HTML document. Before exporting the notebook to html, all of the code cells need to have been run so that reviewers can see the final implementation and output. You can then export the notebook by using the menu above and navigating to \n", "File -> Download as -> HTML (.html). Include the finished document along with this notebook as your submission.
In addition to implementing code, there will be questions that you must answer which relate to the project and your implementation. Each section where you will answer a question is preceded by a 'Question X' header. Carefully read each question and provide thorough answers in the following text boxes that begin with 'Answer:'. Your project submission will be evaluated based on your answers to each of the questions and the implementation you provide.
Note: Code and Markdown cells can be executed using the Shift + Enter keyboard shortcut. Markdown cells can be edited by double-clicking the cell to enter edit mode.
The rubric contains optional "Stand Out Suggestions" for enhancing the project beyond the minimum requirements. If you decide to pursue the "Stand Out Suggestions", you should include the code in this IPython notebook.
In this notebook, you will make the first steps towards developing an algorithm that could be used as part of a mobile or web app. At the end of this project, your code will accept any user-supplied image as input. If a dog is detected in the image, it will provide an estimate of the dog's breed. If a human is detected, it will provide an estimate of the dog breed that is most resembling. The image below displays potential sample output of your finished project (... but we expect that each student's algorithm will behave differently!).

In this real-world setting, you will need to piece together a series of models to perform different tasks; for instance, the algorithm that detects humans in an image will be different from the CNN that infers dog breed. There are many points of possible failure, and no perfect algorithm exists. Your imperfect solution will nonetheless create a fun user experience!
We break the notebook into separate steps. Feel free to use the links below to navigate the notebook.
In the code cell below, we import a dataset of dog images. We populate a few variables through the use of the load_files function from the scikit-learn library:
train_files, valid_files, test_files - numpy arrays containing file paths to imagestrain_targets, valid_targets, test_targets - numpy arrays containing onehot-encoded classification labels dog_names - list of string-valued dog breed names for translating labels%%script false
from sklearn.datasets import load_files
from tensorflow.keras.utils import to_categorical
import numpy as np
from glob import glob
# define function to load train, test, and validation datasets
def load_dataset(path):
data = load_files(path)
dog_files = np.array(data['filenames'])
dog_targets = to_categorical(np.array(data['target']), 133)
return dog_files, dog_targets
# load train, test, and validation datasets
train_files, train_targets = load_dataset('dogImages/train')
print("Done Loading training set")
valid_files, valid_targets = load_dataset('dogImages/valid')
print("Done Loading validation set")
test_files, test_targets = load_dataset('dogImages/test')
print("Done Loading testing set")
# load list of dog names
dog_names = [item[20:-1] for item in sorted(glob("dogImages/train/*/"))]
# print statistics about the dataset
print('There are %d total dog categories.' % len(dog_names))
print('There are %s total dog images.\n' % len(np.hstack([train_files, valid_files, test_files])))
print('There are %d training dog images.' % len(train_files))
print('There are %d validation dog images.' % len(valid_files))
print('There are %d test dog images.'% len(test_files))
# These objects took so much time, so let's pickle them to be used later!!
# make temporary directory
os.mkdir("tmp")
# save to a pickle
def save_object(obj, filename):
with open(filename, 'wb') as fout:
pickle.dump(obj, fout)
save_object(train_files, "tmp/train_files.pickle")
save_object(train_targets, "tmp/train_targets.pickle")
save_object(valid_files, "tmp/valid_files.pickle")
save_object(valid_targets, "tmp/valid_targets.pickle")
save_object(test_files, "tmp/test_files.pickle")
save_object(test_targets, "tmp/test_targets.pickle")
save_object(dog_names, "tmp/dog_names.pickle")
Make sure everything went as expected. The output from the following cell should be:
There are 133 total dog categories.
There are 8351 total dog images.
There are 6680 training dog images.
There are 835 validation dog images.
There are 836 test dog images.
import numpy as np
import pickle
# load some data objects
def load_object(filename):
with open(filename, 'rb') as fin:
obj = pickle.load(fin)
return obj
train_files = load_object("tmp/train_files.pickle")
train_targets = load_object("tmp/train_targets.pickle")
valid_files = load_object("tmp/valid_files.pickle")
valid_targets = load_object("tmp/valid_targets.pickle")
test_files = load_object("tmp/test_files.pickle")
test_targets = load_object("tmp/test_targets.pickle")
dog_names = load_object("tmp/dog_names.pickle")
# print statistics about the dataset
print('There are %d total dog categories.' % len(dog_names))
print('There are %s total dog images.\n' % len(np.hstack([train_files, valid_files, test_files])))
print('There are %d training dog images.' % len(train_files))
print('There are %d validation dog images.' % len(valid_files))
print('There are %d test dog images.'% len(test_files))
There are 133 total dog categories. There are 8351 total dog images. There are 6680 training dog images. There are 835 validation dog images. There are 836 test dog images.
In the code cell below, we import a dataset of human images, where the file paths are stored in the numpy array human_files.
%%script false
import random
random.seed(8675309)
# load filenames in shuffled human dataset
human_files = np.array(glob("lfw/*/*"))
random.shuffle(human_files)
# print statistics about the dataset
print('There are %d total human images.' % len(human_files))
# save the object
save_object(human_files, "tmp/human_files.pickle")
making sure that everything is going as expected... afterr running the following cell, you should get this output:
There are 13233 total human images.
# load object
human_files = load_object("tmp/human_files.pickle")
# print statistics about the dataset
print('There are %d total human images.' % len(human_files))
There are 13233 total human images.
We use OpenCV's implementation of Haar feature-based cascade classifiers to detect human faces in images. OpenCV provides many pre-trained face detectors, stored as XML files on github. We have downloaded one of these detectors and stored it in the haarcascades directory.
In the next code cell, we demonstrate how to use this detector to find human faces in a sample image.
import cv2
import matplotlib.pyplot as plt
%matplotlib inline
# extract pre-trained face detector
face_cascade = cv2.CascadeClassifier('haarcascades/haarcascade_frontalface_alt.xml')
# load color (BGR) image
img = cv2.imread(human_files[3])
# convert BGR image to grayscale
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
# find faces in image
faces = face_cascade.detectMultiScale(gray)
# print number of faces detected in the image
print('Number of faces detected:', len(faces))
# get bounding box for each detected face
for (x,y,w,h) in faces:
# add bounding box to color image
cv2.rectangle(img,(x,y),(x+w,y+h),(255,0,0),2)
# convert BGR image to RGB for plotting
cv_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
# display the image, along with bounding box
plt.imshow(cv_rgb)
plt.show()
Number of faces detected: 1
Before using any of the face detectors, it is standard procedure to convert the images to grayscale. The detectMultiScale function executes the classifier stored in face_cascade and takes the grayscale image as a parameter.
In the above code, faces is a numpy array of detected faces, where each row corresponds to a detected face. Each detected face is a 1D array with four entries that specifies the bounding box of the detected face. The first two entries in the array (extracted in the above code as x and y) specify the horizontal and vertical positions of the top left corner of the bounding box. The last two entries in the array (extracted here as w and h) specify the width and height of the box.
We can use this procedure to write a function that returns True if a human face is detected in an image and False otherwise. This function, aptly named face_detector, takes a string-valued file path to an image as input and appears in the code block below.
# returns "True" if face is detected in image stored at img_path
def face_detector(img_path):
img = cv2.imread(img_path)
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
faces = face_cascade.detectMultiScale(gray)
return len(faces) > 0
Question 1: Use the code cell below to test the performance of the face_detector function.
human_files have a detected human face? dog_files have a detected human face? Ideally, we would like 100% of human images with a detected face and 0% of dog images with a detected face. You will see that our algorithm falls short of this goal, but still gives acceptable performance. We extract the file paths for the first 100 images from each of the datasets and store them in the numpy arrays human_files_short and dog_files_short.
Answer:
human_files_short = human_files[:100]
dog_files_short = train_files[:100]
# Do NOT modify the code above this line.
## TODO: Test the performance of the face_detector algorithm
## on the images in human_files_short and dog_files_short.
acc = sum([face_detector(img_path) for img_path in human_files_short])
print("Accuracy of face_detector() over human files is: {}%".format(acc))
acc = 100-sum([face_detector(img_path) for img_path in dog_files_short])
print("Accuracy of face_detector() over dog files is: {}%".format(acc))
Accuracy of face_detector() over human files is: 100% Accuracy of face_detector() over dog files is: 88%
Question 2: This algorithmic choice necessitates that we communicate to the user that we accept human images only when they provide a clear view of a face (otherwise, we risk having unneccessarily frustrated users!). In your opinion, is this a reasonable expectation to pose on the user? If not, can you think of a way to detect humans in images that does not necessitate an image with a clearly presented face?
Answer:
I think it's fair to ask the user that, but it would be even better if we can accept any image posted by the user. We can detect object in images using cascade classifiers as shown here in this notebook. Or we can use other deep learning algorithms like YOLO.
We suggest the face detector from OpenCV as a potential way to detect human images in your algorithm, but you are free to explore other approaches, especially approaches that make use of deep learning :). Please use the code cell below to design and test your own face detection algorithm. If you decide to pursue this optional task, report performance on each of the datasets.
## (Optional) TODO: Report the performance of another
## face detection algorithm on the LFW dataset
### Feel free to use as many code cells as needed.
In this section, we use a pre-trained ResNet-50 model to detect dogs in images. Our first line of code downloads the ResNet-50 model, along with weights that have been trained on ImageNet, a very large, very popular dataset used for image classification and other vision tasks. ImageNet contains over 10 million URLs, each linking to an image containing an object from one of 1000 categories. Given an image, this pre-trained ResNet-50 model returns a prediction (derived from the available categories in ImageNet) for the object that is contained in the image.
from keras.applications.resnet50 import ResNet50
# define ResNet50 model
ResNet50_model = ResNet50(weights='imagenet')
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version. Instructions for updating: Colocations handled automatically by placer.
When using TensorFlow as backend, Keras CNNs require a 4D array (which we'll also refer to as a 4D tensor) as input, with shape
$$ (\text{nb_samples}, \text{rows}, \text{columns}, \text{channels}), $$where nb_samples corresponds to the total number of images (or samples), and rows, columns, and channels correspond to the number of rows, columns, and channels for each image, respectively.
The path_to_tensor function below takes a string-valued file path to a color image as input and returns a 4D tensor suitable for supplying to a Keras CNN. The function first loads the image and resizes it to a square image that is $224 \times 224$ pixels. Next, the image is converted to an array, which is then resized to a 4D tensor. In this case, since we are working with color images, each image has three channels. Likewise, since we are processing a single image (or sample), the returned tensor will always have shape
The paths_to_tensor function takes a numpy array of string-valued image paths as input and returns a 4D tensor with shape
Here, nb_samples is the number of samples, or number of images, in the supplied array of image paths. It is best to think of nb_samples as the number of 3D tensors (where each 3D tensor corresponds to a different image) in your dataset!
from keras.preprocessing import image
from tqdm import tqdm
def path_to_tensor(img_path):
# loads RGB image as PIL.Image.Image type
img = image.load_img(img_path, target_size=(224, 224))
# convert PIL.Image.Image type to 3D tensor with shape (224, 224, 3)
x = image.img_to_array(img)
# convert 3D tensor to 4D tensor with shape (1, 224, 224, 3) and return 4D tensor
return np.expand_dims(x, axis=0)
def paths_to_tensor(img_paths):
list_of_tensors = [path_to_tensor(img_path) for img_path in tqdm(img_paths)]
return np.vstack(list_of_tensors)
Using TensorFlow backend.
Getting the 4D tensor ready for ResNet-50, and for any other pre-trained model in Keras, requires some additional processing. First, the RGB image is converted to BGR by reordering the channels. All pre-trained models have the additional normalization step that the mean pixel (expressed in RGB as $[103.939, 116.779, 123.68]$ and calculated from all pixels in all images in ImageNet) must be subtracted from every pixel in each image. This is implemented in the imported function preprocess_input. If you're curious, you can check the code for preprocess_input here.
Now that we have a way to format our image for supplying to ResNet-50, we are now ready to use the model to extract the predictions. This is accomplished with the predict method, which returns an array whose $i$-th entry is the model's predicted probability that the image belongs to the $i$-th ImageNet category. This is implemented in the ResNet50_predict_labels function below.
By taking the argmax of the predicted probability vector, we obtain an integer corresponding to the model's predicted object class, which we can identify with an object category through the use of this dictionary.
from keras.applications.resnet50 import preprocess_input, decode_predictions
def ResNet50_predict_labels(img_path):
# returns prediction vector for image located at img_path
img = preprocess_input(path_to_tensor(img_path))
return np.argmax(ResNet50_model.predict(img))
While looking at the dictionary, you will notice that the categories corresponding to dogs appear in an uninterrupted sequence and correspond to dictionary keys 151-268, inclusive, to include all categories from 'Chihuahua' to 'Mexican hairless'. Thus, in order to check to see if an image is predicted to contain a dog by the pre-trained ResNet-50 model, we need only check if the ResNet50_predict_labels function above returns a value between 151 and 268 (inclusive).
We use these ideas to complete the dog_detector function below, which returns True if a dog is detected in an image (and False if not).
### returns "True" if a dog is detected in the image stored at img_path
def dog_detector(img_path):
prediction = ResNet50_predict_labels(img_path)
return ((prediction <= 268) & (prediction >= 151))
Question 3: Use the code cell below to test the performance of your dog_detector function.
human_files_short have a detected dog? dog_files_short have a detected dog?Answer:
### TODO: Test the performance of the dog_detector function
### on the images in human_files_short and dog_files_short.
acc = 100-sum([dog_detector(img_path) for img_path in human_files_short])
print("Accuracy of dog_detector() over human files is: {}%".format(acc))
acc = sum([dog_detector(img_path) for img_path in dog_files_short])
print("Accuracy of dog_detector() over dog files is: {}%".format(acc))
Accuracy of dog_detector() over human files is: 98% Accuracy of dog_detector() over dog files is: 100%
Now that we have functions for detecting humans and dogs in images, we need a way to predict breed from images. In this step, you will create a CNN that classifies dog breeds. You must create your CNN from scratch (so, you can't use transfer learning yet!), and you must attain a test accuracy of at least 1%. In Step 5 of this notebook, you will have the opportunity to use transfer learning to create a CNN that attains greatly improved accuracy.
Be careful with adding too many trainable layers! More parameters means longer training, which means you are more likely to need a GPU to accelerate the training process. Thankfully, Keras provides a handy estimate of the time that each epoch is likely to take; you can extrapolate this estimate to figure out how long it will take for your algorithm to train.
We mention that the task of assigning breed to dogs from images is considered exceptionally challenging. To see why, consider that even a human would have great difficulty in distinguishing between a Brittany and a Welsh Springer Spaniel.
| Brittany | Welsh Springer Spaniel |
|---|---|
![]() |
![]() |
It is not difficult to find other dog breed pairs with minimal inter-class variation (for instance, Curly-Coated Retrievers and American Water Spaniels).
| Curly-Coated Retriever | American Water Spaniel |
|---|---|
![]() |
![]() |
Likewise, recall that labradors come in yellow, chocolate, and black. Your vision-based algorithm will have to conquer this high intra-class variation to determine how to classify all of these different shades as the same breed.
| Yellow Labrador | Chocolate Labrador | Black Labrador |
|---|---|---|
![]() |
![]() |
![]() |
We also mention that random chance presents an exceptionally low bar: setting aside the fact that the classes are slightly imabalanced, a random guess will provide a correct answer roughly 1 in 133 times, which corresponds to an accuracy of less than 1%.
Remember that the practice is far ahead of the theory in deep learning. Experiment with many different architectures, and trust your intuition. And, of course, have fun!
We rescale the images by dividing every pixel in every image by 255.
from PIL import ImageFile
ImageFile.LOAD_TRUNCATED_IMAGES = True
# pre-process the data for Keras
train_tensors = paths_to_tensor(train_files).astype('float32')/255
valid_tensors = paths_to_tensor(valid_files).astype('float32')/255
test_tensors = paths_to_tensor(test_files).astype('float32')/255
100%|██████████| 6680/6680 [01:02<00:00, 106.50it/s] 100%|██████████| 835/835 [00:11<00:00, 74.76it/s] 100%|██████████| 836/836 [00:08<00:00, 97.71it/s]
Create a CNN to classify dog breed. At the end of your code cell block, summarize the layers of your model by executing the line:
model.summary()
We have imported some Python modules to get you started, but feel free to import as many modules as you need. If you end up getting stuck, here's a hint that specifies a model that trains relatively fast on CPU and attains >1% test accuracy in 5 epochs:

Question 4: Outline the steps you took to get to your final CNN architecture and your reasoning at each step. If you chose to use the hinted architecture above, describe why you think that CNN architecture should work well for the image classification task.
Answer:
from keras.layers import Conv2D, MaxPooling2D, GlobalAveragePooling2D
from keras.layers import Dropout, Flatten, Dense
from keras.models import Sequential
from keras.callbacks import ModelCheckpoint
model = Sequential()
### TODO: Define your architecture.
model.add(Conv2D(16, (2, 2), padding='valid', strides=(1,1), activation="relu", input_shape=(224, 224, 3)))
model.add(MaxPooling2D(pool_size=(2,2), strides=(2,2), padding='valid'))
model.add(Conv2D(32, (2, 2), padding='valid', strides=(1,1), activation="relu"))
model.add(MaxPooling2D(pool_size=(2,2), strides=(2,2), padding='valid'))
model.add(Conv2D(64, (2, 2), padding='valid', strides=(1,1), activation="relu"))
model.add(MaxPooling2D(pool_size=(2,2), strides=(2,2), padding='valid'))
model.add(GlobalAveragePooling2D())
model.add(Dense(133, activation='softmax'))
model.summary()
_________________________________________________________________ Layer (type) Output Shape Param # ================================================================= conv2d_1 (Conv2D) (None, 223, 223, 16) 208 _________________________________________________________________ max_pooling2d_2 (MaxPooling2 (None, 111, 111, 16) 0 _________________________________________________________________ conv2d_2 (Conv2D) (None, 110, 110, 32) 2080 _________________________________________________________________ max_pooling2d_3 (MaxPooling2 (None, 55, 55, 32) 0 _________________________________________________________________ conv2d_3 (Conv2D) (None, 54, 54, 64) 8256 _________________________________________________________________ max_pooling2d_4 (MaxPooling2 (None, 27, 27, 64) 0 _________________________________________________________________ global_average_pooling2d_1 ( (None, 64) 0 _________________________________________________________________ dense_1 (Dense) (None, 133) 8645 ================================================================= Total params: 19,189 Trainable params: 19,189 Non-trainable params: 0 _________________________________________________________________
model.compile(optimizer='rmsprop', loss='categorical_crossentropy', metrics=['accuracy'])
Train your model in the code cell below. Use model checkpointing to save the model that attains the best validation loss.
You are welcome to augment the training data, but this is not a requirement.
### TODO: specify the number of epochs that you would like to use to train the model.
epochs = 20
### Do NOT modify the code below this line.
checkpointer = ModelCheckpoint(filepath='saved_models/weights.best.from_scratch.hdf5',
verbose=1, save_best_only=True)
model.fit(train_tensors, train_targets,
validation_data=(valid_tensors, valid_targets),
epochs=epochs, batch_size=20, callbacks=[checkpointer], verbose=1)
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version. Instructions for updating: Use tf.cast instead. Train on 6680 samples, validate on 835 samples Epoch 1/20 6680/6680 [==============================] - 9s 1ms/step - loss: 4.8830 - acc: 0.0093 - val_loss: 4.8658 - val_acc: 0.0156 Epoch 00001: val_loss improved from inf to 4.86582, saving model to saved_models/weights.best.from_scratch.hdf5 Epoch 2/20 6680/6680 [==============================] - 7s 1ms/step - loss: 4.8555 - acc: 0.0139 - val_loss: 4.8338 - val_acc: 0.0204 Epoch 00002: val_loss improved from 4.86582 to 4.83384, saving model to saved_models/weights.best.from_scratch.hdf5 Epoch 3/20 6680/6680 [==============================] - 7s 1ms/step - loss: 4.8066 - acc: 0.0181 - val_loss: 4.7990 - val_acc: 0.0216 Epoch 00003: val_loss improved from 4.83384 to 4.79904, saving model to saved_models/weights.best.from_scratch.hdf5 Epoch 4/20 6680/6680 [==============================] - 8s 1ms/step - loss: 4.7668 - acc: 0.0192 - val_loss: 4.7770 - val_acc: 0.0204 Epoch 00004: val_loss improved from 4.79904 to 4.77703, saving model to saved_models/weights.best.from_scratch.hdf5 Epoch 5/20 6680/6680 [==============================] - 8s 1ms/step - loss: 4.7382 - acc: 0.0228 - val_loss: 4.7557 - val_acc: 0.0192 Epoch 00005: val_loss improved from 4.77703 to 4.75565, saving model to saved_models/weights.best.from_scratch.hdf5 Epoch 6/20 6680/6680 [==============================] - 8s 1ms/step - loss: 4.7120 - acc: 0.0256 - val_loss: 4.7445 - val_acc: 0.0192 Epoch 00006: val_loss improved from 4.75565 to 4.74450, saving model to saved_models/weights.best.from_scratch.hdf5 Epoch 7/20 6680/6680 [==============================] - 8s 1ms/step - loss: 4.6898 - acc: 0.0265 - val_loss: 4.7317 - val_acc: 0.0228 Epoch 00007: val_loss improved from 4.74450 to 4.73169, saving model to saved_models/weights.best.from_scratch.hdf5 Epoch 8/20 6680/6680 [==============================] - 7s 1ms/step - loss: 4.6667 - acc: 0.0286 - val_loss: 4.7010 - val_acc: 0.0240 Epoch 00008: val_loss improved from 4.73169 to 4.70095, saving model to saved_models/weights.best.from_scratch.hdf5 Epoch 9/20 6680/6680 [==============================] - 7s 1ms/step - loss: 4.6409 - acc: 0.0304 - val_loss: 4.7130 - val_acc: 0.0228 Epoch 00009: val_loss did not improve from 4.70095 Epoch 10/20 6680/6680 [==============================] - 7s 1ms/step - loss: 4.6181 - acc: 0.0337 - val_loss: 4.6635 - val_acc: 0.0287 Epoch 00010: val_loss improved from 4.70095 to 4.66351, saving model to saved_models/weights.best.from_scratch.hdf5 Epoch 11/20 6680/6680 [==============================] - 7s 1ms/step - loss: 4.6008 - acc: 0.0391 - val_loss: 4.6590 - val_acc: 0.0383 Epoch 00011: val_loss improved from 4.66351 to 4.65895, saving model to saved_models/weights.best.from_scratch.hdf5 Epoch 12/20 6680/6680 [==============================] - 7s 1ms/step - loss: 4.5817 - acc: 0.0410 - val_loss: 4.6372 - val_acc: 0.0299 Epoch 00012: val_loss improved from 4.65895 to 4.63716, saving model to saved_models/weights.best.from_scratch.hdf5 Epoch 13/20 6680/6680 [==============================] - 7s 1ms/step - loss: 4.5567 - acc: 0.0446 - val_loss: 4.6239 - val_acc: 0.0311 Epoch 00013: val_loss improved from 4.63716 to 4.62388, saving model to saved_models/weights.best.from_scratch.hdf5 Epoch 14/20 6680/6680 [==============================] - 7s 1ms/step - loss: 4.5353 - acc: 0.0476 - val_loss: 4.6076 - val_acc: 0.0347 Epoch 00014: val_loss improved from 4.62388 to 4.60760, saving model to saved_models/weights.best.from_scratch.hdf5 Epoch 15/20 6680/6680 [==============================] - 8s 1ms/step - loss: 4.5112 - acc: 0.0551 - val_loss: 4.5914 - val_acc: 0.0383 Epoch 00015: val_loss improved from 4.60760 to 4.59145, saving model to saved_models/weights.best.from_scratch.hdf5 Epoch 16/20 6680/6680 [==============================] - 7s 1ms/step - loss: 4.4843 - acc: 0.0519 - val_loss: 4.5608 - val_acc: 0.0407 Epoch 00016: val_loss improved from 4.59145 to 4.56077, saving model to saved_models/weights.best.from_scratch.hdf5 Epoch 17/20 6680/6680 [==============================] - 7s 1ms/step - loss: 4.4595 - acc: 0.0566 - val_loss: 4.5643 - val_acc: 0.0371 Epoch 00017: val_loss did not improve from 4.56077 Epoch 18/20 6680/6680 [==============================] - 7s 1ms/step - loss: 4.4301 - acc: 0.0602 - val_loss: 4.5269 - val_acc: 0.0395 Epoch 00018: val_loss improved from 4.56077 to 4.52688, saving model to saved_models/weights.best.from_scratch.hdf5 Epoch 19/20 6680/6680 [==============================] - 7s 1ms/step - loss: 4.4067 - acc: 0.0638 - val_loss: 4.5117 - val_acc: 0.0455 Epoch 00019: val_loss improved from 4.52688 to 4.51165, saving model to saved_models/weights.best.from_scratch.hdf5 Epoch 20/20 6680/6680 [==============================] - 7s 1ms/step - loss: 4.3810 - acc: 0.0656 - val_loss: 4.5142 - val_acc: 0.0395 Epoch 00020: val_loss did not improve from 4.51165
<keras.callbacks.History at 0x7fe56b9cb0b8>
model.load_weights('saved_models/weights.best.from_scratch.hdf5')
Try out your model on the test dataset of dog images. Ensure that your test accuracy is greater than 1%.
# get index of predicted dog breed for each image in test set
dog_breed_predictions = [np.argmax(model.predict(np.expand_dims(tensor, axis=0))) for tensor in test_tensors]
# report test accuracy
test_accuracy = 100*np.sum(np.array(dog_breed_predictions)==np.argmax(test_targets, axis=1))/len(dog_breed_predictions)
print('Test accuracy: %.4f%%' % test_accuracy)
Test accuracy: 5.7416%
bottleneck_features = np.load('bottleneck_features/DogVGG16Data.npz')
train_VGG16 = bottleneck_features['train']
valid_VGG16 = bottleneck_features['valid']
test_VGG16 = bottleneck_features['test']
The model uses the the pre-trained VGG-16 model as a fixed feature extractor, where the last convolutional output of VGG-16 is fed as input to our model. We only add a global average pooling layer and a fully connected layer, where the latter contains one node for each dog category and is equipped with a softmax.
VGG16_model = Sequential()
VGG16_model.add(GlobalAveragePooling2D(input_shape=train_VGG16.shape[1:]))
VGG16_model.add(Dense(133, activation='softmax'))
VGG16_model.summary()
_________________________________________________________________ Layer (type) Output Shape Param # ================================================================= global_average_pooling2d_2 ( (None, 512) 0 _________________________________________________________________ dense_2 (Dense) (None, 133) 68229 ================================================================= Total params: 68,229 Trainable params: 68,229 Non-trainable params: 0 _________________________________________________________________
VGG16_model.compile(loss='categorical_crossentropy', optimizer='rmsprop', metrics=['accuracy'])
checkpointer = ModelCheckpoint(filepath='saved_models/weights.best.VGG16.hdf5',
verbose=1, save_best_only=True)
VGG16_model.fit(train_VGG16, train_targets,
validation_data=(valid_VGG16, valid_targets),
epochs=20, batch_size=20, callbacks=[checkpointer], verbose=1)
Train on 6680 samples, validate on 835 samples Epoch 1/20 6680/6680 [==============================] - 2s 326us/step - loss: 11.9830 - acc: 0.1133 - val_loss: 10.0326 - val_acc: 0.2156 Epoch 00001: val_loss improved from inf to 10.03255, saving model to saved_models/weights.best.VGG16.hdf5 Epoch 2/20 6680/6680 [==============================] - 1s 192us/step - loss: 9.2784 - acc: 0.2963 - val_loss: 9.0633 - val_acc: 0.3222 Epoch 00002: val_loss improved from 10.03255 to 9.06334, saving model to saved_models/weights.best.VGG16.hdf5 Epoch 3/20 6680/6680 [==============================] - 1s 190us/step - loss: 8.4825 - acc: 0.3792 - val_loss: 8.6863 - val_acc: 0.3485 Epoch 00003: val_loss improved from 9.06334 to 8.68632, saving model to saved_models/weights.best.VGG16.hdf5 Epoch 4/20 6680/6680 [==============================] - 1s 189us/step - loss: 8.1104 - acc: 0.4263 - val_loss: 8.5271 - val_acc: 0.3749 Epoch 00004: val_loss improved from 8.68632 to 8.52711, saving model to saved_models/weights.best.VGG16.hdf5 Epoch 5/20 6680/6680 [==============================] - 1s 191us/step - loss: 7.9559 - acc: 0.4522 - val_loss: 8.5814 - val_acc: 0.3605 Epoch 00005: val_loss did not improve from 8.52711 Epoch 6/20 6680/6680 [==============================] - 1s 188us/step - loss: 7.7607 - acc: 0.4750 - val_loss: 8.2902 - val_acc: 0.4012 Epoch 00006: val_loss improved from 8.52711 to 8.29020, saving model to saved_models/weights.best.VGG16.hdf5 Epoch 7/20 6680/6680 [==============================] - 1s 190us/step - loss: 7.6032 - acc: 0.4897 - val_loss: 8.2256 - val_acc: 0.3880 Epoch 00007: val_loss improved from 8.29020 to 8.22555, saving model to saved_models/weights.best.VGG16.hdf5 Epoch 8/20 6680/6680 [==============================] - 1s 189us/step - loss: 7.4928 - acc: 0.5042 - val_loss: 8.0849 - val_acc: 0.4108 Epoch 00008: val_loss improved from 8.22555 to 8.08486, saving model to saved_models/weights.best.VGG16.hdf5 Epoch 9/20 6680/6680 [==============================] - 1s 191us/step - loss: 7.1885 - acc: 0.5232 - val_loss: 7.8774 - val_acc: 0.4180 Epoch 00009: val_loss improved from 8.08486 to 7.87738, saving model to saved_models/weights.best.VGG16.hdf5 Epoch 10/20 6680/6680 [==============================] - 1s 189us/step - loss: 7.1134 - acc: 0.5379 - val_loss: 7.8822 - val_acc: 0.4204 Epoch 00010: val_loss did not improve from 7.87738 Epoch 11/20 6680/6680 [==============================] - 1s 190us/step - loss: 7.0838 - acc: 0.5463 - val_loss: 7.9121 - val_acc: 0.4216 Epoch 00011: val_loss did not improve from 7.87738 Epoch 12/20 6680/6680 [==============================] - 1s 190us/step - loss: 6.9551 - acc: 0.5464 - val_loss: 7.7476 - val_acc: 0.4347 Epoch 00012: val_loss improved from 7.87738 to 7.74764, saving model to saved_models/weights.best.VGG16.hdf5 Epoch 13/20 6680/6680 [==============================] - 1s 192us/step - loss: 6.7774 - acc: 0.5636 - val_loss: 7.7657 - val_acc: 0.4371 Epoch 00013: val_loss did not improve from 7.74764 Epoch 14/20 6680/6680 [==============================] - 1s 188us/step - loss: 6.7479 - acc: 0.5701 - val_loss: 7.6792 - val_acc: 0.4359 Epoch 00014: val_loss improved from 7.74764 to 7.67917, saving model to saved_models/weights.best.VGG16.hdf5 Epoch 15/20 6680/6680 [==============================] - 1s 189us/step - loss: 6.6682 - acc: 0.5723 - val_loss: 7.6003 - val_acc: 0.4455 Epoch 00015: val_loss improved from 7.67917 to 7.60030, saving model to saved_models/weights.best.VGG16.hdf5 Epoch 16/20 6680/6680 [==============================] - 1s 191us/step - loss: 6.5275 - acc: 0.5831 - val_loss: 7.4905 - val_acc: 0.4527 Epoch 00016: val_loss improved from 7.60030 to 7.49048, saving model to saved_models/weights.best.VGG16.hdf5 Epoch 17/20 6680/6680 [==============================] - 1s 196us/step - loss: 6.4121 - acc: 0.5867 - val_loss: 7.4403 - val_acc: 0.4647 Epoch 00017: val_loss improved from 7.49048 to 7.44027, saving model to saved_models/weights.best.VGG16.hdf5 Epoch 18/20 6680/6680 [==============================] - 1s 191us/step - loss: 6.3431 - acc: 0.5981 - val_loss: 7.3984 - val_acc: 0.4611 Epoch 00018: val_loss improved from 7.44027 to 7.39842, saving model to saved_models/weights.best.VGG16.hdf5 Epoch 19/20 6680/6680 [==============================] - 1s 193us/step - loss: 6.3199 - acc: 0.5996 - val_loss: 7.3665 - val_acc: 0.4599 Epoch 00019: val_loss improved from 7.39842 to 7.36650, saving model to saved_models/weights.best.VGG16.hdf5 Epoch 20/20 6680/6680 [==============================] - 1s 190us/step - loss: 6.2496 - acc: 0.6027 - val_loss: 7.3564 - val_acc: 0.4695 Epoch 00020: val_loss improved from 7.36650 to 7.35637, saving model to saved_models/weights.best.VGG16.hdf5
<keras.callbacks.History at 0x7fe56ad2b860>
VGG16_model.load_weights('saved_models/weights.best.VGG16.hdf5')
Now, we can use the CNN to test how well it identifies breed within our test dataset of dog images. We print the test accuracy below.
# get index of predicted dog breed for each image in test set
VGG16_predictions = [np.argmax(VGG16_model.predict(np.expand_dims(feature, axis=0))) for feature in test_VGG16]
# report test accuracy
test_accuracy = 100*np.sum(np.array(VGG16_predictions)==np.argmax(test_targets, axis=1))/len(VGG16_predictions)
print('Test accuracy: %.4f%%' % test_accuracy)
Test accuracy: 47.0096%
from extract_bottleneck_features import *
def VGG16_predict_breed(img_path):
# extract bottleneck features
bottleneck_feature = extract_VGG16(path_to_tensor(img_path))
# obtain predicted vector
predicted_vector = VGG16_model.predict(bottleneck_feature)
# return dog breed that is predicted by the model
return dog_names[np.argmax(predicted_vector)]
%%script false
# download bottleneck features of:
# RestNet-50
!wget https://s3-us-west-1.amazonaws.com/udacity-aind/dog-project/DogResnet50Data.npz -O bottleneck_features/DogResnet50Data.npz
# Inception
!wget https://s3-us-west-1.amazonaws.com/udacity-aind/dog-project/DogInceptionV3Data.npz -O bottleneck_features/DogInceptionV3Data.npz
# Xception
!wget https://s3-us-west-1.amazonaws.com/udacity-aind/dog-project/DogXceptionData.npz -O bottleneck_features/DogXceptionData.npz
You will now use transfer learning to create a CNN that can identify dog breed from images. Your CNN must attain at least 60% accuracy on the test set.
In Step 4, we used transfer learning to create a CNN using VGG-16 bottleneck features. In this section, you must use the bottleneck features from a different pre-trained model. To make things easier for you, we have pre-computed the features for all of the networks that are currently available in Keras:
The files are encoded as such:
Dog{network}Data.npz
where {network}, in the above filename, can be one of VGG19, Resnet50, InceptionV3, or Xception. Pick one of the above architectures, download the corresponding bottleneck features, and store the downloaded file in the bottleneck_features/ folder in the repository.
In the code block below, extract the bottleneck features corresponding to the train, test, and validation sets by running the following:
bottleneck_features = np.load('bottleneck_features/Dog{network}Data.npz')
train_{network} = bottleneck_features['train']
valid_{network} = bottleneck_features['valid']
test_{network} = bottleneck_features['test']
### Obtain bottleneck features from another pre-trained CNN.
bottleneck_features = np.load('bottleneck_features/DogXceptionData.npz')
train_Xception = bottleneck_features['train']
valid_Xception = bottleneck_features['valid']
test_Xception = bottleneck_features['test']
Create a CNN to classify dog breed. At the end of your code cell block, summarize the layers of your model by executing the line:
<your model's name>.summary()
Question 5: Outline the steps you took to get to your final CNN architecture and your reasoning at each step. Describe why you think the architecture is suitable for the current problem.
Answer:
### Defining architecture.
Xception_model = Sequential()
Xception_model.add(GlobalAveragePooling2D(input_shape=train_Xception.shape[1:]))
Xception_model.add(Dense(133, activation='softmax'))
Xception_model.summary()
_________________________________________________________________ Layer (type) Output Shape Param # ================================================================= global_average_pooling2d_1 ( (None, 2048) 0 _________________________________________________________________ dense_1 (Dense) (None, 133) 272517 ================================================================= Total params: 272,517 Trainable params: 272,517 Non-trainable params: 0 _________________________________________________________________
# Compile the model
Xception_model.compile(loss='categorical_crossentropy', optimizer='Adam', metrics=['accuracy'])
Train your model in the code cell below. Use model checkpointing to save the model that attains the best validation loss.
You are welcome to augment the training data, but this is not a requirement.
checkpointer = ModelCheckpoint(filepath='saved_models/weights.best.Xception.hdf5',
verbose=1, save_best_only=True)
Xception_model.fit(train_Xception, train_targets,
validation_data=(valid_Xception, valid_targets),
epochs=20, batch_size=20, callbacks=[checkpointer], verbose=1)
Train on 6680 samples, validate on 835 samples Epoch 1/20 6680/6680 [==============================] - 4s 546us/step - loss: 1.1175 - acc: 0.7325 - val_loss: 0.5376 - val_acc: 0.8335 Epoch 00001: val_loss improved from inf to 0.53761, saving model to saved_models/weights.best.Xception.hdf5 Epoch 2/20 6680/6680 [==============================] - 2s 370us/step - loss: 0.3500 - acc: 0.8924 - val_loss: 0.5131 - val_acc: 0.8443 Epoch 00002: val_loss improved from 0.53761 to 0.51306, saving model to saved_models/weights.best.Xception.hdf5 Epoch 3/20 6680/6680 [==============================] - 3s 377us/step - loss: 0.2278 - acc: 0.9332 - val_loss: 0.4795 - val_acc: 0.8515 Epoch 00003: val_loss improved from 0.51306 to 0.47951, saving model to saved_models/weights.best.Xception.hdf5 Epoch 4/20 6680/6680 [==============================] - 2s 366us/step - loss: 0.1589 - acc: 0.9576 - val_loss: 0.4599 - val_acc: 0.8575 Epoch 00004: val_loss improved from 0.47951 to 0.45994, saving model to saved_models/weights.best.Xception.hdf5 Epoch 5/20 6680/6680 [==============================] - 2s 368us/step - loss: 0.1141 - acc: 0.9714 - val_loss: 0.4824 - val_acc: 0.8515 Epoch 00005: val_loss did not improve from 0.45994 Epoch 6/20 6680/6680 [==============================] - 2s 372us/step - loss: 0.0857 - acc: 0.9805 - val_loss: 0.4575 - val_acc: 0.8527 Epoch 00006: val_loss improved from 0.45994 to 0.45749, saving model to saved_models/weights.best.Xception.hdf5 Epoch 7/20 6680/6680 [==============================] - 3s 382us/step - loss: 0.0671 - acc: 0.9894 - val_loss: 0.4586 - val_acc: 0.8623 Epoch 00007: val_loss did not improve from 0.45749 Epoch 8/20 6680/6680 [==============================] - 3s 391us/step - loss: 0.0541 - acc: 0.9919 - val_loss: 0.4956 - val_acc: 0.8527 Epoch 00008: val_loss did not improve from 0.45749 Epoch 9/20 6680/6680 [==============================] - 3s 402us/step - loss: 0.0449 - acc: 0.9936 - val_loss: 0.4682 - val_acc: 0.8515 Epoch 00009: val_loss did not improve from 0.45749 Epoch 10/20 6680/6680 [==============================] - 3s 387us/step - loss: 0.0361 - acc: 0.9954 - val_loss: 0.4849 - val_acc: 0.8503 Epoch 00010: val_loss did not improve from 0.45749 Epoch 11/20 6680/6680 [==============================] - 3s 379us/step - loss: 0.0311 - acc: 0.9967 - val_loss: 0.4785 - val_acc: 0.8539 Epoch 00011: val_loss did not improve from 0.45749 Epoch 12/20 6680/6680 [==============================] - 2s 363us/step - loss: 0.0280 - acc: 0.9961 - val_loss: 0.4867 - val_acc: 0.8503 Epoch 00012: val_loss did not improve from 0.45749 Epoch 13/20 6680/6680 [==============================] - 2s 359us/step - loss: 0.0215 - acc: 0.9978 - val_loss: 0.5075 - val_acc: 0.8515 Epoch 00013: val_loss did not improve from 0.45749 Epoch 14/20 6680/6680 [==============================] - 2s 360us/step - loss: 0.0208 - acc: 0.9975 - val_loss: 0.4965 - val_acc: 0.8563 Epoch 00014: val_loss did not improve from 0.45749 Epoch 15/20 6680/6680 [==============================] - 2s 357us/step - loss: 0.0175 - acc: 0.9976 - val_loss: 0.5501 - val_acc: 0.8539 Epoch 00015: val_loss did not improve from 0.45749 Epoch 16/20 6680/6680 [==============================] - 2s 364us/step - loss: 0.0160 - acc: 0.9981 - val_loss: 0.5178 - val_acc: 0.8647 Epoch 00016: val_loss did not improve from 0.45749 Epoch 17/20 6680/6680 [==============================] - 2s 360us/step - loss: 0.0142 - acc: 0.9978 - val_loss: 0.5196 - val_acc: 0.8611 Epoch 00017: val_loss did not improve from 0.45749 Epoch 18/20 6680/6680 [==============================] - 2s 363us/step - loss: 0.0131 - acc: 0.9981 - val_loss: 0.5168 - val_acc: 0.8575 Epoch 00018: val_loss did not improve from 0.45749 Epoch 19/20 6680/6680 [==============================] - 2s 361us/step - loss: 0.0120 - acc: 0.9979 - val_loss: 0.5470 - val_acc: 0.8539 Epoch 00019: val_loss did not improve from 0.45749 Epoch 20/20 6680/6680 [==============================] - 2s 358us/step - loss: 0.0114 - acc: 0.9984 - val_loss: 0.5363 - val_acc: 0.8587 Epoch 00020: val_loss did not improve from 0.45749
<keras.callbacks.History at 0x7fe56a3bb2e8>
# Load best weights
Xception_model.load_weights('saved_models/weights.best.Xception.hdf5')
Try out your model on the test dataset of dog images. Ensure that your test accuracy is greater than 60%.
# get index of predicted dog breed for each image in test set
Xception_predictions = [np.argmax(Xception_model.predict(np.expand_dims(feature, axis=0))) for feature in test_Xception]
# report test accuracy
test_accuracy = 100*np.sum(np.array(Xception_predictions)==np.argmax(test_targets, axis=1))/len(Xception_predictions)
print('Test accuracy: %.4f%%' % test_accuracy)
Test accuracy: 85.4067%
| Model | Accuracy |
|---|---|
| VGG16 | 45.5742% |
| ResNet-50 | 81.9378% |
| Inception-V3 | 84.0909% |
| Xception | 85.4067% |
As we can see, the best model is Xception with 85.3% test accuracy.
Write a function that takes an image path as input and returns the dog breed (Affenpinscher, Afghan_hound, etc) that is predicted by your model.
Similar to the analogous function in Step 5, your function should have three steps:
dog_names array defined in Step 0 of this notebook to return the corresponding breed.The functions to extract the bottleneck features can be found in extract_bottleneck_features.py, and they have been imported in an earlier code cell. To obtain the bottleneck features corresponding to your chosen CNN architecture, you need to use the function
extract_{network}
where {network}, in the above filename, should be one of VGG19, Resnet50, InceptionV3, or Xception.
### TODO: Write a function that takes a path to an image as input
### and returns the dog breed that is predicted by the model.
from extract_bottleneck_features import *
def Xception_predict_breed(img_path):
# extract bottleneck features
bottleneck_feature = extract_Xception(path_to_tensor(img_path))
# obtain predicted vector
predicted_vector = Xception_model.predict(bottleneck_feature)
# return dog breed that is predicted by the model
return dog_names[np.argmax(predicted_vector)]
Write an algorithm that accepts a file path to an image and first determines whether the image contains a human, dog, or neither. Then,
You are welcome to write your own functions for detecting humans and dogs in images, but feel free to use the face_detector and dog_detector functions developed above. You are required to use your CNN from Step 5 to predict dog breed.
Some sample output for our algorithm is provided below, but feel free to design your own user experience!

from google.colab.patches import cv2_imshow
### TODO: Write your algorithm.
### Feel free to use as many code cells as needed.
def breed_prediction(img_path):
if face_detector(img_path):
text = "Hello, human!\nYou look like a ...\n"
text+= Xception_predict_breed(img_path)
elif dog_detector(img_path):
text = "Hello, dog!\nYour predicted breed is ...\n"
text+= Xception_predict_breed(img_path)
else:
raise ModuleNotFoundError
print(text)
img = cv2.imread(img_path, 0)
#Display the image
cv2_imshow(img)
In this section, you will take your new algorithm for a spin! What kind of dog does the algorithm think that you look like? If you have a dog, does it predict your dog's breed accurately? If you have a cat, does it mistakenly think that your cat is a dog?
Test your algorithm at least six images on your computer. Feel free to use any images you like. Use at least two human and two dog images.
Question 6: Is the output better than you expected :) ? Or worse :( ? Provide at least three possible points of improvement for your algorithm.
Answer:
face_detector or dog_detector. So, we need to use better techniques.## TODO: Execute your algorithm from Step 6 on
## at least 6 images on your computer.
## Feel free to use as many code cells as needed.
breed_prediction("images/meFace.jpg")
Hello, human! You look like a ... Chinese_crested
breed_prediction("images/sebastian_thrun.jpg")
Hello, human! You look like a ... Canaan_dog
breed_prediction("images/Brittany_02625.jpg")
Hello, dog! Your predicted breed is ... Brittany
breed_prediction("images/Curly-coated_retriever_03896.jpg")
Hello, dog! Your predicted breed is ... Curly-coated_retriever
breed_prediction("images/Labrador_retriever_06449.jpg")
Hello, dog! Your predicted breed is ... Labrador_retriever
breed_prediction("images/Labrador_retriever_06455.jpg")
Hello, dog! Your predicted breed is ... Labrador_retriever
breed_prediction("images/Labrador_retriever_06457.jpg")
Hello, dog! Your predicted breed is ... Labrador_retriever
breed_prediction("images/American_water_spaniel_00648.jpg")
Hello, dog! Your predicted breed is ... Curly-coated_retriever